Google launched Gemini, its latest generative AI chatbot that guarantees to be higher than ChatGPT, alongside a splashy demo video showcasing its capabilities. A Google developer weblog submit reveals that the viral showcase was edited, sped up, and used completely different prompts than those proven within the video.
“For the needs of this demo, latency has been lowered and Gemini outputs have been shortened for brevity,” mentioned Google within the Gemini demo video description, however that doesn’t describe the total quantity of modifying right here.
The discharge of Gemini was an enormous advertising and marketing second for Google’s synthetic intelligence group, which has confronted strain from buyers to meet up with OpenAI. Google’s important AI chatbot Bard has lagged behind ChatGPT and different fashions in 2023, so Gemini wanted to assist Google return to the chopping fringe of the AI dialog – and it did. Nonetheless, Gemini’s demo video appears to have been an over-exaggeration of its precise capabilities. Google pushed again the unique launch of Gemini as a result of Sundar Pichai deemed it wasn’t prepared, and now we’re seeing why.
The demo video appeared as if Gemini works with speech, video visuals, and hand gestures, however that’s not how Google bought these solutions. Behind the scenes, Gemini required extra cautious, longer prompts with textual content and pictures to get these solutions within the video, based on a Google weblog submit. “Fingers-on with Gemini” appeared as if the AI was reacting as a real-time assistant with video and voice capabilities, however that was deceptive, and customers won’t have these options anytime quickly. The video was seemingly filmed with Gemini Extremely, essentially the most superior model of Google’s AI that received’t be launched till at the least subsequent yr.
Google’s claims that Gemini is healthier than ChatGPT appear deceptive as nicely, as Bloomberg factors out. Google touted a mannequin of Gemini that doesn’t exist but (Extremely, coming 2024) in opposition to outdated fashions from OpenAI (ChatGPT-4, from March 2023). The benchmark doesn’t examine to ChatGPT-4 Turbo, OpenAI’s most superior AI mannequin that was launched final month. Google’s mannequin that’s obtainable as we speak, Gemini Professional, doesn’t beat ChatGPT-4 in Huge Multitask Language Understanding – it loses by over 7%.
As customers bought their first likelihood to make use of Gemini Professional on Wednesday, many felt the expertise lacked the luster of Google’s advertising and marketing marketing campaign. Gemini Professional appears to ‘hallucinate’ so much, a typical downside the place chatbots confidently give false solutions. One person posted on X that Gemini Professional couldn’t reply who received finest actor on the 2023 Oscars. One other person discovered that Gemini Professional informed them to “Google it” when requested questions on Israel and Gaza.